A spatially seamless, daily FY-3B soil moisture dataset based on GSP model

The Fengyun-3B (FY-3B) satellite, equipped with the Microwave Radiation Imager (MWRI), provides an effective way for monitoring soil moisture (SM) at regional to global scales, serving as an important data source within the suite of SM products. However, the FY-3B SM product suffers from substantial missing data due to orbital gaps, which considerably limits its practical applicability. To address this challenge, we propose a parallel spatiotemporal reconstruction framework, termed the GSP (multi-scale Gated Convolution–residual Shifted Window Transformer Parallel) model, which integrates local fine-grained features with global semantic representations. The GSP model fully exploits the spatiotemporal characteristics of FY-3B SM and leverages complementary modules to enhance feature representation. Based on the GSP framework, we generated a global, spatially continuous, daily FY-3B SM dataset spanning from 12 July 2011 to 19 August 2019. This work provides methodological guidance for large-scale reconstruction of FY-3B SM data, while the resulting dataset offers valuable support for research in fields such as soil science and hydrology.